Aurora forecasting, explained by people who had to build one.

Writing an aurora forecast engine forces you to confront how much of the popular explanation of this subject is wrong, or right in a way that stops one step before the part that matters. The Kp index is described everywhere as a measure of aurora strength. It is not. It is a planetary average of geomagnetic disturbance over three hours, built from magnetometer stations spread around the world, and turning it into an answer for one person standing in one field requires several more steps that most apps and most articles quietly skip.

This section is where we write those steps down. It exists because we needed it: when we started, the available material split into three unhelpful categories. NOAA and the Geophysical Institute publish the real data with real rigour, but they publish it as instrument output for people who already know what they are looking at. Tour operators and travel sites write beautifully readable explanations attached to something they want to sell you, and the optimism shows. And Reddit fills the gap in between with genuine expertise scattered through threads you have to already know how to find.

We sell an app, so we are not neutral either. What we can do is be useful to you even if you never install it — which is why the free tools on this site run the same calculations the app runs, why the comparison pages say plainly where competitors beat us, and why the flagship article in this section is an honest account of how accurate aurora forecasts actually are, broken down by lead time. That article makes our own product look more limited than the marketing copy of any competitor. It is also true.

Where to start, depending on what you want

If you want the short version: aurora visibility is a chain of four conditions, and the weakest link decides your night. There has to be enough geomagnetic activity to push the auroral oval down to your latitude. You have to be close enough to that oval in corrected geomagnetic latitude, which is not the latitude on your map. Your sky has to be genuinely dark, which above the Arctic Circle is impossible for months at a time. And it has to be clear, which no forecast model can help you with. Every disappointing aurora night is one of those four links, and usually it is the last one.

If you are planning a trip, start with the season and timing pieces. They answer the questions that actually change a booking: which months work at your destination, why the equinoxes are favoured, what magnetic midnight means for the hour you should be outside, and why the number of nights you have available matters more than any forecast. One clear night in a week beats a perfect Kp on a single-night visit, and that is the most useful planning advice in this entire subject.

If you already chase and you want the parts that separate a good night from a wasted drive, go to the solar wind article. Kp is a lagging summary; the leading indicators are the interplanetary magnetic field orientation, the solar wind speed and the density, measured by spacecraft roughly an hour upstream of Earth. A strongly southward Bz at moderate speed is a better sign than a fast stream pointing north, and no app that shows only a Kp number can tell you which one you are looking at.

If you want to photograph it, read the camera pieces before you travel rather than in the dark with cold hands. The single most common failure is not exposure but focus: achieving true infinity focus at night is genuinely difficult and the failure is not obvious on a small screen. The second most common is expectation. A phone photograph of a faint display looks far greener than the display did, because a camera integrates light over seconds while dark-adapted human vision is close to colourblind. That gap surprises almost every first-time viewer, and knowing about it in advance is the difference between a good night and a let-down.

Every article here is written to be read on its own, so there is repetition between them by design — you should not have to read five pages to understand one. Where a piece depends on a concept explained properly somewhere else, it links there rather than compressing it into a misleading sentence. Corrections are welcome and we make them: if something here is wrong, tell us and we will fix it and say that we did.

Five things almost everyone gets wrong

The first is treating Kp as a local measurement. It is a planetary index averaged over three hours, which means a sharp substorm lasting twenty minutes can produce a superb display inside a three-hour block that reports a modest number, and a high number can describe activity that was concentrated on the other side of the planet while you were outside seeing nothing. Kp tells you how disturbed the magnetosphere was globally. It does not tell you what happened over your head.

The second is reading the coloured oval map as a probability for your garden. The OVATION model shows modelled emission overhead at each point, and because the aurora sits around a hundred kilometres up you can see it from well outside the coloured region by looking toward the horizon. People regularly conclude they have no chance because the green blob stops a few hundred kilometres north of them, when in fact a display on the poleward horizon was perfectly visible.

The third is assuming the aurora zone follows geographic latitude. It follows the geomagnetic field, whose pole is offset from the geographic pole and moves measurably from year to year. That offset pushes the zone south over North America and north over Siberia, which is why the northern United States can see displays that a Russian city at the same map latitude will not.

The fourth is expecting the naked eye to match the photographs. It usually will not, except during strong events. Most of the images that set expectations were made with long exposures, and the honest description of a moderate display is a pale, slowly shifting arc that looks grey-green until it brightens.

The fifth is over-trusting long-range forecasts. Beyond about an hour, aurora forecasting is probabilistic in a way that resists precision, because the decisive variable — the direction of the magnetic field embedded in the arriving solar wind — is only measured once that wind passes the spacecraft sitting upstream of Earth. Any product promising a specific night three weeks out is describing a recurrence pattern, not a forecast.